Introduction to Anaphora and Anaphora
Types of Anaphora
Process of Anaphora Resolution
Ruslan Mitkov, School of Languages and European
Studies University of Wolverhampton, Stafford Street, UK
◦ ANAPHORA RESOLUTION: THE STATE OF THE ART
◦ Outstanding Issues in Anaphora Resolution
Anaphora in Etymology
◦ Ancient Greek : Anaphora = anajora (Anajora)
ana (Ana) back in an upward direction
jora (Jora ) the act of carrying back upstream
◦ The Empress hasn't arrived yet but she should be
here any minute.
The Empress (NP) Antecedent
Empress (N) is NOT the antecedent !
Coreferent Both The Empress and she refer to the
same REAL WORLD ENTITY
◦ when the “anaphor” precedes the “antecedent”
◦ Because she was going to the post office, Julie was
asked to post a small parcel
Anaphora Resolution(AR) is the process of
determining the antecedent of an anaphor.
◦ Anaphor – The reference that points to the
◦ Antecedent –The entity to which anaphor refers
Needed to derive the “Correct Interpretation”
of a text
Is a complicated problem in NLP !
Natural Language Interfaces
when the anaphor and more than one of the
preceding (or following) entities (usually
noun phrases) have the same referent and
are therefore pairwise coreferential
◦ System has to determine antecedents of anaphors
◦ Identify all coreference CHAINS
◦ the task of AR is considered successful, if any of
the preceding entities in the coreferential chain is
identified as an antecedent.
◦ The most widespread type of anaphora
◦ Realized by anaphoric pronouns
Computational Linguists from many different
countries attended the tutorial. They took extensive
◦ Not all pronouns in English are anaphoric
It is raining.
It is non-anaphoric
Definite noun phrase anaphora
◦ when the antecedent is referred by a definite
noun phrase representing either same concept
(repetition) or semantically close concepts (e.g.
Computational Linguists from many different countries
attended the tutorial. The participants found it hard to
cope with the speed of the presentation.
◦ One-anaphora is the case when the anaphoric
expression is realized by a "one" noun phrase.
If you cannot attend a tutorial in the morning, you can
go for an afternoon one.
Some are easier than others.
The red ones went in the side pocket.
◦ Referring to an antecedent which is in the same
sentence as the anaphor
◦ Referring to an antecedent which is in a different
sentence from that of the anaphor
Possibilities: identifying anaphors which have
noun phrases, verb phrases, clauses,
sentences or even paragraphs/discourse
segments as antecedents
Most of the AR systems deal with : identifying
anaphors which have noun phrases as their
All noun phrases (NPs) preceding an anaphor are
initially regarded as potential candidates for
◦ Search scope
Most approaches look for NPs in the current and
Antecedents which are 17 sentences away from the
anaphor have already been reported !
◦ Uses Charniak’s parser
◦ Input Minimally Associated XML format (MAS-XML)
◦ Output in XML
◦ John Hopkins University Tool for AR
◦ Uses Charniak’s parser
◦ Comes up with Machine learning model
◦ Can take Input as XML. Output XML
◦ Built on Mitkov principle
◦ Uses certain linguistic rules
◦ If your data has noises, don’t use it.
◦ Working on Lipin and Leass rules (1994)
◦ Uses Charniak’s parser
CherryPicker: A Coreference Resolution Tool
Reconcile - Coreference Resolution Engine
Why BART ?
◦ Open Source
◦ It is beautiful (Beautiful Anaphora Resolution
◦ It works !
Where do we stand today?
◦ Domain-specific and linguistic knowledge are not
◦ Knowledge-poor anaphora strategies have
By emergence of cheaper and more reliable corpus-
based NLP tools such as POS taggers, Sallow parsers,
and other NLP resources (ontologies)
◦ Use of modern approaches
Accuracy of Pre-processing is still TOO low
◦ Morphological Analysis / POS tagging -- fair
◦ Name Entity Reorganization – still a challenge
Best performing NER 96 % when trained and tested on news about a
SPECIFIC topic and 93 % when trained on news about topic and tested
on news about other topic
◦ Unknown word recognition
◦ NP extraction – have a long way to go
NP chunking 90%-93% recall and precision
◦ Gender recognition – still a challenge
◦ Identification of pleonastic pronouns
Best Accuracy in robust parsing of unrestricted texts = 87 %
How accurate an AR algorithm we make, it won’t perform
well until the accuracy of pre-pressing improves
Majority of Anaphora resolution systems do
not operate in fully automatic mode.
Fully automatic AR is more difficult than
◦ 54.65 % in MARS
◦ It was as high as 90 % for perfectly analyzed inputs
Need for Annotated Corpora
◦ Needed for training Machine Learning Algorithms or
◦ Corpora annotated with anaphoric or coreferential
links are not widely available
◦ Annotation tool
◦ Annotation scheme
Evaluation in Anaphora Resolution
Factors (constraints and preferences)
◦ Mutually dependent
Most people still work mainly on pronoun
Service to Research Community
◦ Sharing software, experience, data
◦ There are just 3 demos for AR ;)
“NLP in general is very difficult but after
working hard on Anaphora Resolution we
have learned that it is particularly difficult”~
"A pessimist sees the difficulty in every
opportunity; an optimist sees the opportunity
in every difficulty”~Winston Churchill
“All we have to do is work more and HOPE for
SLOW but STEADY progress. We just have to
be PATIENT !” ~ Mitkov
◦ AR in multi-person dialogue
Haven’t found any paper specific to Tutoring
Massimo Poesio Slides: Anaphora Resolution for Practical Tasks, University of
Offenbar haben Sie einen Ad-Blocker installiert. Wenn Sie SlideShare auf die Whitelist für Ihren Werbeblocker setzen, helfen Sie unserer Gemeinschaft von Inhaltserstellern.
Sie hassen Werbung?
Wir haben unsere Datenschutzbestimmungen aktualisiert.
Wir haben unsere Datenschutzbestimmungen aktualisiert, um den neuen globalen Regeln zum Thema Datenschutzbestimmungen gerecht zu werden und dir einen Einblick in die begrenzten Möglichkeiten zu geben, wie wir deine Daten nutzen.
Die Einzelheiten findest du unten. Indem du sie akzeptierst, erklärst du dich mit den aktualisierten Datenschutzbestimmungen einverstanden.